Cancer Biology: A New RING to Wnt Signaling

Abstract

The β-catenin-dependent Wnt signaling pathway has key roles in embryonic development and adult tissues, and mutations in the pathway underlie the development of colorectal and other cancers. Recent studies highlight the role of two transmembrane RING finger ubiquitin ligases in modulating Wnt signaling at the cell surface as well as the role of and implications for mutations of the RING finger proteins in cancer.

Cancers often harbor defects in evolutionarily conserved pathways regulating embryonic development and cell fate specification in adult tissues. The Wnt pathway is a classic example of the phenomenon. The cysteine-rich Wnt proteins are roughly 40 kDa secreted molecules, and Wnts function as close-range signals for their context-specific effects on varied cell phenotypes (1,2). The actions of Wnts on target cells include changes in gene expression and cell polarization and directed migration, via engagement of distinct downstream molecules (1). For the “canonical” or β-catenin-dependent Wnt pathway, Wnts bind to a receptor complex, composed of a low-density lipoprotein-related protein 5 or 6 (LRP5/6) molecule and a Frizzled (Fz) protein. LRP5/6 have a single membrane-spanning domain, and Fz proteins are seven-transmembrane (7-TM) molecules. There is much complexity in Wnt-Fz interactions, with 19 Wnt and 10 Fz molecules (2). In the absence of activating Wnt signals at the cell surface, a “destruction complex” for uncomplexed β-catenin in the cell is assembled, consisting of glycogen synthase kinase 3β, the Axin and APC (adenomatous polyposis coli) tumor suppressor proteins, and other factors. This complex phosphorylates β-catenin in its amino (N)-terminal domain. The phosphorylated β-catenin is ubiquitinated, and then degraded by the proteasome. Wnt activation of the Fz-LRP5/6 complex inhibits β-catenin degradation, apparently via inhibition of β-catenin ubiquitination (2,3) (Fig. 1A). Newly synthesized β-catenin protein can thus accumulate in the cell. In the nucleus, β-catenin can bind to T cell factor (TCF) transcription factors along with other proteins, and the β-catenin/TCF complexes modulate transcription of selected genes with roles in cell fate, proliferation, and other processes (1,2).

Figure 1.

Model for Wnt regulation of the free pool of β-catenin. Left: In the absence of an activating Wnt ligand, β-catenin is phosphorylated at multiple serine/threonine residues in its amino-terminal domain by the ‘destruction complex’, consisting of Axin, APC, GSK3β, and casein kinase 1α (CK1α). The amino-terminally phosphorylated β-catenin is recognized by a ubiquitin ligation complex that contains βTrCP, and the ubiquitinated β-catenin is degraded by the proteasome. The ubquitination and degradation of β-catenin allows the destruction complex to bind and phosphorylate additional β-catenin molecules. Right: In the presence of an activating Wnt ligand, which bridges the Fz–LRP5/6 proteins and brings about phosphorylation of cytoplasmic LRP5/6 sequences by CK1a, the destruction complex is recruited to the phosphorylated cytoplasmic domain of LRP5/6. While the destruction complex can still bind and phosphorylate β-catenin, ubiquitination by βTrCP is blocked, allowing newly synthesized β-catenin to accumulate in the cell and activate transcription in the nucleus, via interactions with TCF proteins and various co-factors, including the p300/CBP histone acetyltransferases, BCL9, and Pygopus (Pygo), among others.

Mutations in the β-catenin-dependent Wnt signaling pathway contribute especially to colorectal cancer (CRC). About 90% of CRCs have somatic mutations affecting certain canonical Wnt pathway factors (4,5). More than 50% of hepatocellular carcinomas (HCCs) have mutations in the canonical Wnt pathway (6), as do significant subsets of other cancer types (1,2,7). Mutational mechanisms include inactivation of the APC protein in most CRCs (1,2,4,5) or the AXIN1 protein in some HCCs (6), or activating (oncogenic) mutations in key phosphorylation sites in β-catenin’s N-terminal domain in HCCs and other cancer types (1,2,7). A major consequence of the mutations is that β-catenin is constitutively stabilized in the absence of Wnt signals, with resultant altered transcription of β-catenin/TCF-regulated genes.

Several secreted and transmembrane proteins modulate Wnt ligand signaling through the Fz-LRP5/6 receptor. Wnt ligand binding inhibitors include secreted Frizzled-related proteins (sFRPs) and the secreted Wnt inhibitor factor 1 (WIF1) (2). The Dickkopf (DKK) and Sclerostin (SOST) secreted proteins interfere Wnt-simulated Fz-LRP5/6 interactions, and the APCDD1 (APC down-regulated 1) transmembrane protein interferes with Wnt binding to LRP5/6 (2). Two distinct protein families function through the Fz-LRP5/6 complex to enhance Wnt signaling. Norrin binds directly to certain Fz proteins (e.g., Fz4) to activate canonical Wnt signaling independent of Wnts (2). In contrast, the four R(oof plate-specific)-spondin proteins are secreted, vertebrate-specific factors that enhance signaling via Fz-LRP5/6 complexes, but only in the presence of Wnt ligands (2).

Until recently, a major unresolved issue was how R-spondins enhanced Wnt ligand-dependent signaling. The first clues were provided by demonstrations that R-spondins bind to leucine-rich, G-protein coupled receptor 5 (Lgr5) proteins (8–10). Lgr5, and the related proteins Lgr4 and Lgr6, are 7-TM receptors with similarly to the G-protein coupled hormone receptors, such as the receptor for thryoid-stimulating hormone (2,11). Prior efforts had shown Lgr5 gene expression was activated by Wnt/β-catenin/TCF-dependent signaling (2,11). More significantly, Lgr5 is expressed by intestinal stem and progenitor cells, such as the so-called crypt base columnar (CBC) cells in intestinal crypts, and Lgr5-expressing intestinal cells had tissue stem cell properties in vivo and in intestinal organoid models. Lgr5 expression was also found to mark stem cells in other tissues (11). Lgr6 expression also seems to mark presumptive tissue stem cells in certain organs, while Lgr4 expression is more broad and not likely restricted to stem cells (11). The findings that R-spondins bound to Lgr5 family members were obviously intriguing. However, the means by which the R-spondin-Lgr4/5/6 interactions might enhance Wnt signaling through the Fz-LRP5/6 co-receptor complex remained unclear. Two recent papers in Nature have provided powerful insights into how R-spondins likely enhance Wnt signaling (12,13), and the findings have important implications for the cancer field.

Both papers took advantage of gene expression profiling as a starting point. Prior efforts had demonstrated that multiple proteins encoded by genes activated by β-catenin/TCF functioned as negative feedback regulators of Wnt signaling, including AXIN2, DKK1, NKD1, APCDD1, and WIF1 (1,2). Hao and colleagues sought to define genes whose expression was tightly correlated with AXIN2 expression and that might also regulate the Wnt pathway (12). Gene expression for two closely related transmembrane RING finger proteins with E3 ubiquitin ligase function, known as ZNRF3 and RNF43, was linked to β-catein function and largely paralleled that of AXIN2, including in CRCs (12). The same two genes drew the attention of Koo and colleagues in their analysis of genes expressed in Lgr5-expressing CBCs (13). The two papers diverge in terms of the specific approaches and models used to explore ZNRF3 and RNF43 function. However, both papers arrive at similar conclusions; namely, that ZNRF3 and RNF43 regulate the stability and levels of cell surface Fz and LRP5/6 proteins via ubiquitination and then internalization and lysomal degradation of the receptor components. Both groups propose that ZNRF3 and RNF43 function, akin to AXIN2 and some other Wnt targets, as negative feedback regulators of Wnt signaling.

The paper from Hao and colleagues also provides data indicating that R-spondins function to stabilize Wnt receptor levels on the cell surface through the ability of an Lgr5 family member and a ZNFR3/RNF43 member to bind R-spondins (12). R-spondin binding to this Lgr/TM-RING receptor complex leads to the membrane clearance of the TM-RING protein, with resultant increased levels of the Fz-LRP5/6 receptor complex and enhanced Wnt signaling (12) (Fig. 1B). The paper from Koo and colleagues explores the roles of RNF43 and ZNRF3 as presumptive tumor suppressor proteins, demonstrating that conditional inactivation of both Rnf43 and Znrf3 in murine intestinal epithelium led to marked expansion of the proliferative zone of the crypt and the formation of epithelial tumors (13). The tumors resembled intestinal adenomas and were composed chiefly of cells found at the base of normal intestinal crypts – i.e., the CBCs (expressing Lgr5 and other stem cell markers) and Paneth cells (13). Paneth cells are a specialized cell type at the crypt base in normal small intestine, which may play a role in establishing and maintaining a stem cell niche, perhaps due to their supply of Wnt, Notch, and/or epidermal growth factor signals to CBCs. Using intestinal organoids generated from Rnf43/Znrf3 double-mutant epithelial cells, Koo and colleagues demonstrated the mutant organoids no longer required Rspo1 supplementation, but were highly dependent on locallly supplied Wnts, such as Paneth cell-derived Wnt3.

Hao and colleagues noted that RNF43 inactivating mutations were found in some pancreas tumors (14,15), and Koo and colleagues noted that two CRC cell lines with activating β-catenin mutations (and not APC inactivation) had inactivating RNF43 mutations (16). In the HCT116 CRC cell line with β-catenin and RNF43 defects, Koo and colleagues illustrated how RNF43 inactvation may allow for Wnt ligands to enhance β-catenin/TCF signaling beyond that achieved by a β-catenin mutation alone (13). RNF43 is inactivated in various cancers, as 10% of bile duct cancers had RNF43 mutations in a recent study (17). Some prior published results may also be clarified by the findings in these Nature papers. The Aaronson group had previously reported the HCT116 CRC line had Wnt ligand dependence for β-catenin/TCF signaling (18), and their group’s recent report of over-expression of LRP5/6 protein in some sarcomas in the absence of demonstrable defects in the genes for LRP5/6 could be due in some cases to inactivation of RNF43 (19).

Whether RNF43 and ZNRF3 negatively regulate all or nearly all Fz-LRP5/6 co-receptor complexes or only a subset is uncertain The identities of other proteins that might be regulated by RNF43/ZNRF3 is also unknown. Whether all R-spondins can remove the TM-RING complex from the cell surface remains to be determined. Cell context and variable expression patterns for RNF43 and ZNRF3 might offer some explanation for why combined somatic inactivation of Rnf43 and Znrf3 was required to instigate intestinal tumorigenesis in the mouse, yet only the RNF43 gene seems to be somatically mutated in human cancers. The basis for why various other negative feedback regulators of the Wnt pathway, such as NKD1, DKK1, and APCDD1 are not recurrently mutated in cancer is obscure, though baseline redundancy for some of their functions in Wnt signaling may be part of the answer. Data addressing some of these issues will likely be forthcoming. Moreover, given the potential opportunities for targeting Wnt signaling in cancers where RNF43 (and/or possibly ZNRF3) are defective, there will likely be considerable enthusiasm for new strategies and reagents to target Wnt ligands and receptors directly in certain subsets of cancer.